Re: Changing the state of data checksums in a running cluster

Hello Michael and Tomas,

24.08.2026 04:25, Michael Paquier wrote:
> Is the coverage provided by debug_parallel_query useful to have in
> this context?

Probably not, but 203 of 224 buildfarm animals have
'debug_parallel_query = regress' in their configs. Only 2 animals test
checksum_extended (partridge and porpoise) for now, but if someone tried
to enable this extra test on an animal from the majority, they would hit
the failure, I guess.

>> queries start enough parallel workers to exhaust max_worker_processes,
>> and so we fail to start the checksum worker.
>>
>> I've modified the TAP tests to add
>>
>> max_worker_processes = 16
>> max_parallel_workers = 16
>>
>> and with that change it passes just fine.

Yeah, these settings work for me too.

> One option would be debug_parallel_query = off to shave the number of
> workers spawned.

Yes, there are a couple of tests that do this already, namely:
009_log_temp_files.pl
002_worker_terminate.pl -- though that config adjustment must be
  redundant now, after 283c5fb22.

Best regards,
Alexander